visolex/VITHSD

VITHSD（越南语针对性仇恨言论检测）包含10,000条越南语社交媒体评论，这些评论针对五个目标类别进行了仇恨程度的标注：个人、群体、宗教/信仰、种族/民族和政治。每个目标类别都按照三个等级（0=无仇恨，1=冒犯，2=仇恨）进行标注。这个统一版本将所有划分合并为一个CSV文件，并增加了一个`type`列，表示训练集、验证集或测试集。

VITHSD (Vietnamese Targeted Hate Speech Detection) contains 10,000 Vietnamese social-media comments annotated for hate toward five target categories: individual, groups, religion/creed, race/ethnicity, and politics. Each target is labeled on a 3-point scale (0 = no hate, 1 = offensive, 2 = hateful). In this unified version, all splits are combined into one CSV with an extra type column indicating train / dev / test.